Article ID: CJ-15-0129
Background: Tricuspid regurgitation (TR) is a common echocardiographic finding that has been related to adverse outcome under various clinical scenarios. Nevertheless, evidence supporting its prognostic value in heart failure (HF) is scarce, and, in most cases, contradictory. We evaluated the association of TR grade with 1-year all-cause mortality in acute HF (AHF).
Methods and Results: We included 1,842 consecutive patients admitted for AHF. Mean age was 72.8±11.3 years, 51% were female and 45.5% had LVEF <50%. The severity of TR was graded in non-TR, mild (1), moderate (2), moderate-severe (3) and severe (4). At 1-year follow-up, 370 patients (20.1%) had died. In patients with LVEF ≥50%, a significant and positive association between TR severity and mortality was noted. Indeed, the HR for mortality for TR 3 and 4 vs. no TR/TR 1 were as follows: hazard ratios (HR), 1.68; 95% confidence intervals (95% CI): 1.08–2.60, P=0.02; and HR, 2.87; 95% CI: 1.61–5.09, P<0.001, respectively. In contrast, no association between TR grade and mortality (P=0.650) was observed in patients with LVEF <50% (P-value for interaction=0.033).
Conclusions: A differential prognostic effect of TR severity on 1-year mortality was observed for LVEF HF status. The association was significant only in patients with LVEF ≥50%, with increasing mortality risk as TR became more severe.
Tricuspid regurgitation (TR) is a common echocardiographic finding.1 In patients with heart failure (HF), TR is explained as a functional consequence of the dilatation of the right ventricular (RV) annulus and tricuspid leaflet tethering in patients with RV pressure and/or volume overload. Following this scheme, TR becomes part of a positive feedback cycle where its progression leads to further RV dilatation and dysfunction.2 Observational studies in the field of HF, and conducted more than 10 years ago, noted an associated poor prognosis with significant TR.3,4 Nowadays, there is a resurgence of interest on the potential of such an association under various clinical scenarios.5 A recent study by Neuhold et al sheds light on the significance of TR in chronic HF patients, by reporting a significant association only in patients with reduced ejection fraction (HFrEF).6 Indeed, TR was an independent predictor only in patients with mild-to-moderate left ventricular systolic dysfunction (LVSD), but not in those with severe LVSD. Unfortunately, the literature is scarce on the prognostic effect of TR in HF with preserved ejection fraction (HFpEF), a situation that accounts for nearly 50% of the current admissions for acute HF (AHF) in clinical practice.7 This fact contrasts with the growing interest about the importance of right-heart parameters, such as RV dysfunction and/or pulmonary hypertension (PH), in the pathophysiology, clinical expression, and prognosis of patients with HFpEF.8–12 Therefore, we evaluated the association between functional TR and 1-year all-cause mortality in a non-selected cohort of patients with AHF, and tested for any differential effect of HF type, namely HFrEF and HFpEF.
Editorial p ????
We prospectively included a consecutive cohort of 1,957 patients who were admitted for AHF to the cardiology department of a tertiary-care teaching hospital in Spain (Hospital Clinico Universitario de Valencia) from January 2004 to August 2013. AHF was defined according to current Clinical Practice Guidelines.13 By design, patients with primary tricuspid disease (n=5) or severe mitral rheumatic stenosis (n=46) were excluded. In addition, patients with a final diagnosis of acute coronary syndrome (n=26), pneumonia (n=18), pulmonary thromboembolism (n=8), or those patients who died before echocardiography assessment (n=12) were also excluded. During index hospitalization, data on demographics, medical history, vital signs, 12-lead electrocardiogram, laboratory, echocardiographic parameters, and drug utilization were routinely recorded using pre-established registry questionnaires. Treatment with angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, β-blockers, aldosterone antagonist, anticoagulants, diuretics, and other therapeutic strategies were individualized following established guidelines that were operating at the time the patient was included in the registry.
EchocardiographyIn all patients, 2-D echocardiogram was performed during index hospitalization (96±24 h after admission) using the left lateral decubitus position. Two commercially available systems were used throughout the study: Agilent Sonos 5500 and ie33 (Philips, MA, USA). No patients were on mechanical ventilation at the time of echocardiography. All images were recorded with second-harmonic at the time of end expiration. LV ejection fraction (LVEF) was calculated using the biplane Simpson method. HFrEF and HFpEF were defined as LVEF <50% or ≥50%, respectively. LV diastolic dysfunction was evaluated using transmitral flow Doppler velocities and tissue Doppler imaging-derived mitral annulus velocities. Pulmonary artery systolic pressure (PASP) was estimated by measuring the maximum continuous Doppler-derived velocity of the TR jet, and used to estimate the RV to atrial pressure gradient during systole using the modified Bernoulli equation (4v2), where v is the velocity of the TR jet. Then, this gradient was added to the estimated right atrial pressure to calculate the RV systolic pressure which, in the absence of pulmonary stenosis, is equal to PASP. Even though RV systolic function was not routinely assessed in the entire cohort, the tricuspid annular plane systolic excursion (TAPSE), in contrast, was estimated in 748 patients. TAPSE was obtained by M-mode recordings from the apical 4-chamber view with the cursor placed at the free wall side of the annulus. RV systolic dysfunction was defined as TAPSE <16 mm.14
According to current recommendations, the severity of TR was assessed using an integrated multi-parametric score that includes qualitative and semi-quantitative parameters.15,16 Color-flow Doppler was examined in the apical 4-chamber, parasternal short- and long-axis, and subcostal views. Quantitative measures of the TR jet using the proximal isovelocity surface area method was available in <5% of patients, so it was not used in this analysis. TR severity grade was assigned by the cardiac sonographer and confirmed by another study investigator blinded to clinical data. The TR severity score consists of 4 categories: 1, mild; 2, moderate; 3, moderate-severe; and 4, severe. All regurgitation jets with a vena contracta ≥7 mm and/or systolic flow reversal in hepatic veins were considered to be severe (grade 4). Grade ≥3 was labeled as significant TR.
OutcomeThe primary endpoint of the study was 1-year all-cause mortality with time beginning the day the patient was admitted to the hospital. Secondary endpoint was 1-year cardiovascular mortality. Cause of death was considered as non-cardiovascular if a specific non-cardiovascular cause was identified. Otherwise, cardiovascular etiology of death was considered and included sudden death, progressive HF death, deaths attributable to other cardiovascular causes (such as myocardial infarction, stroke, etc) and unknown cause of death. The survival status was ascertained, in most cases, by reviewing the electronic medical records, although corroborated by an investigator who was blinded to TR and LVEF results.
EthicsThe study was prospectively designed, conformed to the principles outlined in the Declaration of Helsinki, and approved by the institutional local review ethics committee. All patients gave informed consent.
Statistical AnalysisContinuous variables are expressed as mean±SD or median (IQR). Discrete variables are summarized as percentages. Baseline characteristics were compared among TR categories separately in patients with LVEF <50% and LVEF ≥50%. The association between TR with 1-year all-cause mortality was evaluated on Cox regression analysis, and the results expressed as hazard ratios (HR) with 95% confidence intervals (95% CI). For evaluating cardiovascular mortality a Cox regression model adjusted for competing events was used. Candidate covariates were chosen based on previous medical knowledge; then, a backward stepwise selection was performed. The discriminative ability, calibration and proportionality assumption were assessed with Harrell’s C-statistics, the Groennesby and Borgan test, and the Schoenfeld residuals, respectively. Final multivariate analysis (model 1) included all the following covariates assessed during index admission: age, gender, prior HF admissions, last New York Heart Association functional class III/IV under stable conditions, ischemic etiology, LVEF <50% interacting with systolic blood pressure, blood urea nitrogen, hemoglobin, plasma N-terminal pro-B-type natriuretic peptide (NT-proBNP), plasma carbohydrate antigen 125 (CA125), treatment with β-blockers and LVEF <50% interacting with TR grade. Detailed multivariate model including all the covariates and their estimates is given in Table S1. The Harrell’s C-statistics and the Groennesby and Borgan test (χ2) of the final multivariate analysis were 0.742 and 0.391, respectively. In sensitivity analysis PASP and TAPSE were added as covariates, and the use of a different threshold for LVEF (<45% vs. ≥45%; model 3) was also evaluated.
Two-sided P<0.05 was considered to be statistically significant for all analyses. All survival analyses were performed using STATA 13.1 (StataCorp, College Station, TX, USA).
Mean age was 72.8±11.3 years, 51% were female, 35.5% had ischemic etiology, 45.5% had LVEF <50% and 45.6% were previously admitted for AHF. The distribution of patients according TR severity was as follows: 1,117 patients (60.6%) had some degree of TR (grade 1, 33.3%; grade 2, 17.3%; grade 3, 7.4%; and grade 4, 2.7%). Of note, the distribution of TR severity was similar among the 2 groups with regard to LVEF (<50%, HFrEF vs. ≥50%, HFpEF; Figure 1). Baseline patient characteristics according to LVEF (HFrEF vs. HFpEF) are given in Table S2.
Distribution of tricuspid regurgitation (TR) grade with regard to left ventricular ejection fraction (LVEF) status. P=0.826.
Tables 1,2 summarize the baseline characteristics of the study sample, according to TR severity and LVEF status (HFrEF and HFpEF, respectively). Overall, there was a positive association between TR grade and known surrogates of HF severity in both populations, although this assumption was less ostensible in HFpEF. In both groups, patients with higher TR severity were older, had higher prevalence of peripheral edema, were more likely to have ischemic heart disease, atrial fibrillation, left atrial enlargement, and higher median NT-proBNP and CA125. Likewise, they also had lower mean blood pressure, hemoglobin and TAPSE. We found remarkable the fact that TR severity was related to greater comorbidity in HFrEF, but not in HFpEF. Likewise, in HFpEF patients, PASP, TAPSE and NT-proBNP were related to higher TR grade, but the magnitude of these associations was lower than observed in those with HFrEF (Tables 1,2).
| No TR (n=327) |
TR grade 1 (n=289) |
TR grade 2 (n=143) |
TR grade 3 (n=57) |
TR grade 4 (n=22) |
P-value for trend |
|
|---|---|---|---|---|---|---|
| Demographic and medical history | ||||||
| Age (years) | 71±11 | 70±13 | 72±12 | 70±15 | 65±15 | 0.845 |
| Female | 108 (33.0) | 109 (37.7) | 39 (27.3) | 21 (36.8) | 4 (18.2) | 0.334 |
| First HF admission | 193 (59.0) | 158 (54.7) | 69 (48.2) | 25 (43.9) | 10 (45.4) | 0.005 |
| Hypertension | 263 (80.4) | 226 (78.2) | 96 (67.1) | 35 (61.4) | 13 (59.1) | <0.001 |
| Diabetes mellitus | 160 (48.9) | 132 (45.7) | 69 (48.2) | 21 (36.8) | 8 (36.4) | 0.121 |
| Dyslipidemia | 178 (54.4) | 152 (52.6) | 71 (49.6) | 20 (35.1) | 8 (36.4) | 0.006 |
| Current smoker | 60 (18.3) | 45 (15.6) | 18 (12.6) | 9 (15.8) | 9 (13.6) | 0.192 |
| Ischemic heart disease | 173 (52.9) | 124 (42.9) | 48 (33.6) | 29 (50.9) | 8 (36.4) | 0.006 |
| Charlson index† | 2 (2) | 2 (2) | 2 (2) | 2 (2) | 3 (2) | 0.019 |
| Peripheral edema | 156 (47.7) | 158 (54.7) | 105 (73.4) | 48 (84.2) | 18 (81.8) | 0.001 |
| Vital signs | ||||||
| Heart rate (beats/min) | 103±25 | 100±26 | 98±27 | 93±24 | 99±27 | 0.001 |
| SBP (mmHg) | 152±36 | 147±34 | 132±27 | 127±28 | 124±18 | <0.001 |
| DBP (mmHg) | 87±21 | 84±20 | 77±17 | 74±22 | 77±16 | <0.001 |
| ECG | ||||||
| QRS >120 ms | 131 (40.1) | 115 (39.8) | 71 (49.6) | 24 (42.1) | 7 (31.8) | 0.470 |
| Atrial fibrillation | 74 (22.6) | 93 (32.2) | 67 (46.8) | 28 (49.1) | 12 (54.5) | <0.001 |
| Laboratory | ||||||
| Hemoglobin (g/dl) | 13.1±1.9 | 12.9±1.9 | 12.8±2.0 | 12.1±1.7 | 12.1±2.0 | <0.001 |
| Sodium (mEq/L) | 139±4 | 139±4 | 138±4 | 136±5 | 136±5 | <0.001 |
| NT-proBNP (pg/ml)† | 3,314 (3,958) | 4,513 (6,611) | 5,297 (7,395) | 6,045 (7,903) | 10,937 (19,217) | <0.001 |
| CA125 (U/ml)† | 50.4 (99.5) | 71.0 (117.6) | 100.5 (146.2) | 131.0 (261.9) | 116.2 (135.7) | <0.001 |
| Serum creatinine (mg/dl) | 1.30±0.62 | 1.26±0.57 | 1.42±0.64 | 1.38±0.55 | 1.51±0.77 | 0.003 |
| BUN (mg/dl) | 57.4±29.6 | 56.7±29.8 | 65.6±31.3 | 68.7±33.7 | 84.7±43.9 | <0.001 |
| eGFR (ml·min−1·1.73 m−2) | 57.4±29.6 | 56.7±29.8 | 65.5±31.3 | 68.7±33.7 | 84.7±43.9 | 0.016 |
| Echocardiography | ||||||
| LVEF (%) | 37±8 | 35±8 | 35±9 | 35±9 | 30±9 | <0.001 |
| LAD (mm) | 41±6 | 43±8 | 45±6 | 49±9 | 49±9 | <0.001 |
| TAPSE (mm) | 18±3 | 18±3 | 17±2 | 16±3 | 15±3 | <0.001 |
| PASP (mmHg) | – | 39±10 | 49±11 | 51±13 | 51±13 | <0.001 |
| Treatment at discharge | ||||||
| β-blockers | 212 (64.8) | 224 (77.5) | 98 (68.5) | 37 (64.9) | 14 (63.6) | 0.727 |
| ACEI or ARB | 240 (73.4) | 215 (74.4) | 92 (64.3) | 40 (70.2) | 19 (86.4) | 0.545 |
| MRA | 118 (36.1) | 136 (47.1) | 61 (42.7) | 31 (54.4) | 10 (45.4) | 0.013 |
| Loop diuretics | 319 (97.5) | 281 (97.2) | 140 (97.9) | 57 (100) | 21 (95.4) | 0.677 |
Data given as n (%), mean±SD or †median (IQR). ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BUN, blood urea nitrogen; CA125, antigen carbohydrate 125; DBP, diastolic blood pressure; ECG, electrocardiogram; eGFR, estimated glomerular filtration rate; HF, heart failure; MRA, mineralocorticoid receptor blocker; NT-proBNP, N-terminal pro-B-type natriuretic peptide; LAD, left atrial diameter; LVEF, left ventricular ejection fraction; PASP, pulmonary arterial systolic pressure; SBP, systolic blood pressure; TAPSE, tricuspid annular plane systolic excursion; TR, tricuspid regurgitation.
| No TR (n=398) |
TR grade 1 (n=324) |
TR grade 2 (n=175) |
TR grade 3 (n=79) |
TR grade 4 (n=28) |
P-value for trend |
|
|---|---|---|---|---|---|---|
| Demographic and medical history | ||||||
| Age (years) | 74±10 | 75±10 | 76±9 | 74±11 | 73±9 | 0.334 |
| Female | 235 (59.0) | 212 (65.4) | 124 (70.9) | 64 (81.0) | 24 (85.7) | <0.001 |
| First HF admission | 224 (56.3) | 192 (56.3) | 96 (54.9) | 30 (38.0) | 10 (35.7) | 0.004 |
| Hypertension | 341 (85.7) | 265 (81.8) | 139 (79.4) | 58 (73.4) | 23 (82.1) | 0.011 |
| Diabetes mellitus | 188 (47.2) | 138 (42.6) | 58 (33.1) | 28 (35.4) | 11 (39.3) | 0.003 |
| Dyslipidemia | 195 (49.0) | 160 (49.4) | 71 (40.6) | 33 (41.8) | 12 (42.9) | 0.067 |
| Current smoker | 38 (9.5) | 25 (7.7) | 9 (5.1) | 3 (3.8) | 2 (7.1) | 0.040 |
| Ischemic heart disease | 135 (33.9) | 91 (28.1) | 31 (17.7) | 8 (10.1) | 6 (21.4) | <0.001 |
| Charlson index† | 1 (2) | 1.5 (2) | 1 (2) | 1 (1) | 1 (2) | 0.991 |
| Peripheral edema | 204 (51.3) | 217 (67.0) | 119 (68.0) | 66 (83.5) | 22 (78.6) | <0.001 |
| Vital signs | ||||||
| Heart rate (beats/min) | 100±31 | 99±31 | 100±31 | 90±25 | 97±39 | 0.099 |
| SBP (mmHg) | 159±37 | 154±35 | 145±31 | 136±29 | 137±29 | <0.001 |
| DBP (mmHg) | 84±21 | 81±19 | 78±19 | 76±17 | 76±13 | <0.001 |
| ECG | ||||||
| QRS >120 ms | 92 (23.1) | 64 (19.7) | 47 (26.9) | 24 (30.4) | 5 (17.9) | 0.339 |
| Atrial fibrillation | 149 (37.4) | 162 (50.0) | 116 (66.3) | 57 (72.1) | 23 (82.1) | <0.001 |
| Laboratory | ||||||
| Hemoglobin (g/dl) | 12.4±1.8 | 11.9±1.9 | 12.1±1.8 | 12.0±2.0 | 12.3±2.0 | 0.014 |
| Sodium (mEq/L) | 139±4 | 139±5 | 139±6 | 138±5 | 138±5 | 0.274 |
| NT-proBNP (pg/ml)† | 2,412 (2,592) | 2,832 (3,467) | 3,314 (3,590) | 3,125 (4,506) | 3,314 (4,406) | <0.001 |
| CA125 (U/ml)† | 36.2 (66.0) | 46.5 (84.4) | 67.3 (113.6) | 76.2 (140.8) | 127.5 (133.4) | <0.001 |
| Serum creatinine (mg/dl) | 1.24±0.59 | 1.19±0.56 | 1.17±0.55 | 1.18±0.54 | 1.27±0.69 | 0.147 |
| BUN (mg/dl) | 59.8±30.7 | 57.2±26.4 | 59.9±31.8 | 62.9±30.6 | 75.2±58.0 | 0.442 |
| eGFR (ml·min−1·1.73 m−2) | 61.7±25.9 | 65.0±40.0 | 62.3±25.2 | 61.4±26.7 | 58.7±25.9 | 0.903 |
| Echocardiography | ||||||
| LVEF (%) | 62±8 | 62±7 | 62±8 | 62±8 | 62±8 | 0.947 |
| LAD (mm) | 40±7 | 43±7 | 45±8 | 50±10 | 52±9 | <0.001 |
| TAPSE (mm) | 20±3 | 19±4 | 17±5 | 18±5 | 18±3 | <0.001 |
| PASP (mmHg) | – | 42±13 | 52±14 | 64±19 | 60±17 | <0.001 |
| Treatment at discharge | ||||||
| β-blockers | 223 (56.0) | 216 (66.7) | 100 (57.1) | 41 (51.9) | 14 (50.0) | 0.538 |
| ACEI or ARB | 274 (68.8) | 207 (63.9) | 114 (65.1) | 52 (65.8) | 19 (67.9) | 0.467 |
| MRA | 56 (14.1) | 61 (18.8) | 40 (22.9) | 22 (27.8) | 13 (46.4) | <0.001 |
| Loop diuretics | 386 (97.0) | 309 (95.4) | 166 (94.9) | 77 (97.5) | 28 (100) | 0.965 |
Data given as n (%), mean±SD or †median (IQR). Abbreviations as in Table 1.
At 1-year follow-up, 370 patients (20.1%) had died. A total of 288 (77.8%) were registered as cardiovascular deaths. In the whole sample, an almost linear association between TR severity and mortality rates was found: 18.3%, 16.5%, 22.0%, 32.4% and 44% for no TR, grade 1, 2, 3 and 4, respectively. When the sample was stratified according to LVEF status, TR severity was associated with mortality in both groups (HFrEF and HFpEF), although this effect was more evident and uniformly distributed along the entire follow-up in HFpEF (Figure 2).
Kaplan-Meier curves for 1-year all-cause mortality according to tricuspid regurgitation (TR) severity. (A) Heart failure with reduced ejection fraction; (B) heart failure with preserved ejection fraction.
On multivariate analysis, this differential prognostic effect of TR with regard to LVEF status was confirmed (omnibus P-value for interaction=0.033). Breaking down the interaction, we found a proportional and significant increase in risk when moving from no TR/TR grade 1 to grade 4 in HFpEF but not in HFrEF (Table 3). Indeed, compared with patients with no TR/TR grade 1, those with TR grades 3 and 4 had a 68% and 187% increased risk of mortality, respectively (Table 3).
| Cox models | HR (95% CI) | P-value | HR (95% CI) | P-value | P-value for interaction |
|---|---|---|---|---|---|
| All-cause mortality | |||||
| Model 1 | |||||
| LVEF <50% | LVEF ≥50% | 0.033* | |||
| TR grade 2 vs. no TR/TR grade 1 | 0.80 (0.53–1.21) | 0.295 | 1.17 (0.81–1.69) | 0.409 | 0.180 |
| TR grade 3 vs. no TR/TR grade 1 | 0.95 (0.56–1.60) | 0.849 | 1.68 (1.08–2.60) | 0.020 | 0.100 |
| TR grade 4 vs. no TR/TR grade 1 | 0.84 (0.39–1.80) | 0.655 | 2.87 (1.61–5.09) | <0.001 | 0.011 |
| Model 2 (adjusting for PASP) | |||||
| LVEF <50% | LVEF ≥50% | 0.039* | |||
| TR grade 2 vs. no TR/TR grade 1 | 0.78 (0.51–1.20) | 0.254 | 1.13 (0.77–1.67) | 0.525 | 0.179 |
| TR grade 3 vs. no TR/TR grade 1 | 0.92 (0.53–1.58) | 0.758 | 1.60 (0.99–2.58) | 0.054 | 0.110 |
| TR grade 4 vs. no TR/TR grade 1 | 0.81 (0.37–1.77) | 0.602 | 2.74 (1.50–5.01) | 0.001 | 0.012 |
| Model 3 | |||||
| LVEF <45% | LVEF ≥45% | 0.041* | |||
| TR grade 2 vs. no TR/TR grade 1 | 0.66 (0.41–1.07) | 0.102 | 1.23 (0.88–1.71) | 0.226 | 0.036 |
| TR grade 3 vs. no TR/TR grade 1 | 0.83 (0.46–1.50) | 0.544 | 1.67 (1.11–2.52) | 0.013 | 0.058 |
| TR grade 4 vs. no TR/TR grade 1 | 0.96 (0.42–2.18) | 0.919 | 2.20 (1.26–3.84) | 0.006 | 0.098 |
| Cardiovascular mortality† | |||||
| LVEF <50% | LVEF ≥50% | 0.087* | |||
| TR grade 2 vs. no TR/TR grade 1 | 0.93 (0.60–1.44) | 0.753 | 1.39 (0.90–2.14) | 0.140 | 0.206 |
| TR grade 3 vs. no TR/TR grade 1 | 1.23 (0.69–2.22) | 0.485 | 2.35 (1.44–3.83) | 0.001 | 0.095 |
| TR grade 4 vs. no TR/TR grade 1 | 0.84 (0.33–2.15) | 0.717 | 2.86 (1.37–5.94) | <0.001 | 0.044 |
*Omnibus test P-value for interaction. Model 1 and 3 included the following set of covariates: age, gender, prior HF admission, New York Heart Association functional class III/IV, ischemic etiology, LVEF <50% interacting with SBP, BUN, hemoglobin, NT-proBNP, CA125, treatment with β-blockers and LVEF <50% interacting with TR grade. Model 2 adjusted for the same subset of covariates as model 1 plus PASP (non-measurable, normal PASP [PASP ≤35 mmHg], mild PH [PASP 36–45 mmHg], moderate PH [PASP 46–60 mmHg] and severe PH [PASP >60 mmHg]). †Adjusting for the same set of covariates as in models 1 and 3 plus non-cardiovascular death as a competing event. CI, confidence interval; HR, hazard ratio; PH, pulmonary hypertension. Other abbreviations as in Table 1.
On sensitivity analysis, using LVEF=45% as the threshold to define the 2 categories (HFpEF and HFrEF), this differential prognostic effect persisted (Table 3).
In a different sensitivity analysis in which PASP was added to the multivariate model, the main finding was still present. In this model, PASP categories were independently associated with risk of mortality, as long as it was not co-adjusted by TR severity. When both variables were introduced simultaneously, only TR grade was retained in the final model. If PASP was forced in the multivariate model, TR severity remained a significant predictor of mortality risk in HFpEF, but not in HFrEF (model 2; Table 3). Of note, the prognostic value of TR was not significantly different across PASP >60 mmHg (omnibus P-value for interaction=0.968).
In another sensitivity analysis, where TAPSE was added as an additional covariate (using only the 748 patients with data), the mortality risk was again differentially affected by the severity of TR and according to LVEF status (omnibus P-value for interaction=0.075). In this analysis, TR was not related to mortality in HFrEF; in contrast, in HFpEF, the HR for the comparison between TR grades 3 and 4 vs. no TR/TR grade 1 were as follows: HR, 2.07; 95% CI: 1.01–4.27, P=0.048; and HR, 3.65; 95% CI: 1.41–9.46, P=0.008, respectively.
With regard to cardiovascular mortality, a differential prognostic effect of TR grade with regard to LVEF status was also obtained. A significant and positive association between TR and mortality was found in HFpEF but not in HFrEF (Table 3).
The major finding of this study is that, in patients with AHF, the prognostic effect of functional TR differed with regard to HFrEF and HFpEF. In patients with HFpEF, TR was independently related to 1-year survival. In contrast, in HFrEF, TR was related to survival on univariate analysis, but this association disappeared in a multivariate context.
TR and Prognosis in HFAlthough TR is a common echocardiographic finding,1 its importance has been traditionally neglected. Some studies conducted more than a decade ago reported an association between significant functional TR and adverse prognosis in chronic HF.3,4 These studies, however, included only patients with HFrEF in a stable clinical condition. In a heterogeneous cohort of 5,223 patients undergoing echocardiography at the Palo Alto Veterans Affairs Health Care System, Nath et al reported a lower 1-year survival for patients with moderate to severe TR, regardless of PASP or LVEF status.17 The drawback of these early studies is that they lacked adjustment for the currently established risk factors, such as NT-proBNP and renal indices, and that they were not fully representative of the contemporary population of HF. Interestingly, there is a renewed interest on the impact of functional TR under various clinical scenarios.5,18,19 Recently, Neuhold et al studied the prognostic impact of TR in 576 patients with chronic HF and predominantly LVSD (only 17% of patients had HFpEF).6 They reported that TR was significantly related to long-term mortality only in patients with LVEF ≥35%, whereas in severely depressed LVEF patients, TR had no significant impact on outcome. The present results not only are partially in line with these results, but further expanded the TR differential effect on mortality to a large and representative sample of HF patients in the acute setting. Moreover, and supported by the present results, we postulate that it is LVEF status, and not necessarily the severity of HF, that explains the differential impact of TR on survival.
Unraveling the Differential Prognostic Effect of TR Across LVEFHFrEF and HFpEF are 2 different syndromes with well-defined differences in clinical and pathophysiological dimensions.20 In HFrEF, the development of functional TR and its severity are traditionally explained as a deleterious response of the RV and the pulmonary circulation to the elevation in left-side pressures.2 In this scheme of thinking, the severity of TR is mainly envisioned as reflecting the severity of the underlying LVSD, which in turn causes compensatory changes in venous and arterial pulmonary pressure and RV hemodynamics.21 Indeed, in patients with HFrEF, we found a strong association between surrogates of HF severity, such as higher natriuretic peptides, renal dysfunction, prior admissions, lower LVEF or higher PASP, and TR severity score. In contrast, this association was less obvious in HFpEF (Table 2). Within this theoretical framework, it does appear plausible that TR plays a more important and direct prognostic role in HFpEF, but not in HFrEF, in which any potential prognostic effect is mostly mediated through surrogates for LV dysfunction and its consequences.
HFpEF constitutes a heterogeneous syndrome in which cardiac and non-cardiac abnormalities play a crucial pathophysiological role. The most common cardiac and vascular abnormalities involve ventricular and vascular stiffness and the subsequent impaired relaxation. In addition, several extracardiac abnormalities, such as obesity, pulmonary diseases, atrial fibrillation, hypertension, anemia, physical deconditioning or frailty play a pivotal role in the pathophysiology and progression of the disease.22 Thus, and in contrast to HFrEF, most of the clinical features in HFpEF are only partially in line with the left-side dysfunction.23 Indeed, the overlap between cardiac and extracardiac abnormalities reshape the final disease phenotypes.22,24 Most of the mentioned “partners” of LV diastolic dysfunction in HFpEF are also well-established risk factors for PH and for the development/progression of RV dysfunction.10,12,25 Along this line of thinking, the reactive or pre-capillary component of PH is more prevalent in HFpEF than in HFrEF.26 Clearly, this pathogenetic component is not the mere product of the underlying severity of the left-side disease, and is increasingly recognized as a crucial determinant of the HFpEF phenotype.9 We believe that this “ventricular disconnection” found in HFpEF might partially explain the differential prognostic effect between these 2 conditions. Thus, whereas in HFrEF TR severity seems to be mainly determined by the severity of LVSD, in HFpEF TR severity may capture many of the prognostic burdens of the associated conditions causing PH and RV dysfunction, which, in the end, overwhelm the contribution of the left-side disease. In HFpEF, the RV is especially sensitive to pressure overload, showing heightened-RV load sensitivity and diastolic RV stiffness.10 Good to remember are the facts that, in HFpEF, RV contractility is impaired in 20–50% of patients,9 and that PH or RV systolic dysfunction are better correlated with prognosis than the left-heart systolic or diastolic function parameters,10,27 similar to what has been observed in primary pulmonary arterial hypertension.28 This combination of factors has made the isolation and better definition of the role of functional TR in HF, extremely challenging. Related to this is the fact that the presence of significant TR (grade ≥3) was more predictive of survival than the magnitude of echocardiography-derived PH. They are all highly related parameters, given that significant TR is related to increasing PASP.29,30 Nonetheless, TR has theoretical advantages over PASP as a risk factor. First, PASP cannot be estimated when TR is not present. Moreover, in the setting of severe TR, echocardiography-derived estimation of PASP have been shown to be largely inaccurate,14 raising the possibility that some patients could be misdiagnosed using such an estimation. In summary and according to the present results, significant functional TR in HFpEF can be a sensitive and useful parameter to provide meaningful prognostic information by reflecting the final pathway of PH and RV dysfunction.
TR as a Causative Factor for Progression of HFpEFWhether TR may have a role as a causative factor for disease progression in HFpEF or act as an epiphenomenon is another issue that needs to be clarified in further studies. There is cumulative evidence that significant TR leads to RV dilatation and dysfunction.5 When RV fails, RV diastolic pressure rises, and can even lead to a shift in the interventricular septum to LV.2 This reduction in LV size can produce an increase in LV diastolic pressures, a fact that may especially decrease cardiac output in patients with a small and stiff LV chamber, the hallmark of HFpEF.13 These enhanced ventricular interactions have been recently studied by Andersen et al, who showed that patients with TR grade ≥3 and preserved LVEF had increased right atrial pressure combined with inadequate LV diastolic filling, and a consequently impaired exercise tolerance.31 Furthermore, TR is a recognized factor leading to greater systemic venous congestion. Indeed, TR was related to traditional surrogates of congestion in the present study (peripheral edema, NT-proBNP or CA12532). Likewise, in previous studies, functional TR severity has also been correlated with liver congestion markers.33 Furthermore, TR recently emerged as a recognized risk factor predicting the degree of renal dysfunction/worsening renal function in patients with AHF.34
Study LimitationsThis study has the inherent limitations of a single-center observational study, in which the potential for some hidden bias and unmeasured confounders precludes the drawing of a causal statement. TAPSE was not available in every patient, which precluded a formal prognostic comparison between RV systolic function parameters. We are also aware, however, that the accuracy of TAPSE as a surrogate of RV systolic function is limited when significant TR is present.35 The same limitation, however, applies to the peak systolic tricuspid annular velocity35 and other echocardiographic parameters such as fractional area change due to RV load dependency. The complex geometry of the RV makes it difficult to assess systolic function for almost every 2-D echocardiographic parameter.14 NT-proBNP and other biomarkers were not obtained simultaneously to the echocardiographic data. Left-sided mitral and aortic regurgitation severities were not uniformly collected, precluding evaluation of their association with outcome and how they could affect the present results. Finally, echocardiograms were not reviewed by an independent core laboratory external to the investigators. As strengths, we would like to note the fact that this design and the data obtained merely reflect daily clinical practice, and that all echocardiography was conducted by experienced cardiologists dedicated to cardiac imaging.
TR is related to 1-year survival in patients with AHF. This association strongly depends on LVEF HF status. Indeed, only in patients with LVEF ≥50% was the association significant, with increasing mortality risk as TR became more severe.
This work was supported in part by grants from Instituto de Salud Carlos III, Red de Investigación Cardiovascular, Programa 7 (RD12/0042/0010) FEDER; PI11/02323 and Prometeo/2013/007.
The authors have no conflicts of interest to disclose.
Supplementary File 1
Table S1. Multivariate indicators of all-cause 1-year mortality
Table S2. Baseline subject characteristics vs. LVEF (HFpEF vs. HFrEF)
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-15-0129
